Abstract/Summary

Mixed-phase layer clouds are radiatively important and their correct representation in numerical models of the atmosphere is needed for both weather forecasts and climate prediction. In particular, midlevel mixed-phase layer clouds (altocumulus) are often poorly predicted. Here the representation of altocumulus cloud in five operational models and the ERA-Interim reanalysis is evaluated using ground-based remote sensors. All models are found to underestimate the supercooled liquid water content by at least a factor of 2. The models with the most sophisticated microphysics (separate prognostic variables for liquid and ice) had least supercooled liquid of all models, though they could simulate the correct liquid-over-ice structure of individual clouds. To investigate the reasons for the lack of predicted supercooled liquid water, a single-column model (EMPIRE) was developed incorporating the relevant physical processes for altocumulus cloud. The supercooled liquid water was found to be the most sensitive to factors that significantly affect the glaciation rate, including aspects of the ice microphysics formulation, as well as the model vertical resolution. Using observations to improve the ice particle size distribution formulation and the parametrization of ice cloud fraction also lead to a significant increase in supercooled liquid water in the simulated clouds. The study highlights the main parameterized processes that need careful attention in large-scale models in order to adequately represent the liquid phase in mixed-phase layer clouds. In Part 2, the reason for the sensitivity to vertical resolution is investigated and a new parameterization for models with coarse vertical resolution is proposed.